Science1 publisher2 min readPublished
Basecamp raises $140M to push model-designed recombinases toward in vivo cell therapy
The company says the money will train the next generation of its EDEN models. The published evidence so far is one January bioRxiv preprint, co-authored with Nvidia, whose venture arm also put money into the round.
The Scientist · Science desk

What happened
- Basecamp Research says it has closed an oversubscribed $140 million Series C to train the next generation of its EDEN models and apply them to in vivo cell therapies for several diseases.
- EDEN, the company's biological foundation model, is meant to design long and complex DNA sequences along with the large serine recombinases that deliver those sequences precisely.
- Anthropic invested through its Anthology Fund, a $100 million vehicle financed by Menlo Ventures, and Nvidia came in through its venture arm NVentures.
Compiled by The ScientistSomething wrong?How this is made
Why it matters
- constraint The money buys model training, and the recombinase results stop at human cells in culture, so the cost argument for treating patients in vivo has no in-body efficiency number attached to it yet.
- exposure A partner weighing EDEN cannot treat the January preprint as outside validation of the round, because researchers from one of the investors helped write it.
- capability If a model can generate working insertion enzymes from a 30-base-pair prompt, enzyme design stops being the limit on payload size, and the limiting step becomes getting that enzyme into the right cells.
- precedent Investing in a biology startup, hosting its models and co-authoring its papers is now a working template for how model builders back the companies that feed them data.
A large serine recombinase does the insertion step. Basecamp says EDEN designs the long DNA sequence and the enzyme that places it [3], and with Nvidia the company has applied that to programmable gene insertion, putting large therapeutic sequences at precise locations in the human genome [13].
The published evidence is a single preprint, posted January 12 on bioRxiv by researchers from Basecamp, Nvidia, Microsoft and two research institutions [14]. Prompted on only 30 base pairs of DNA from outside its training data, EDEN reached an overall functional hit rate of 63.2 percent across diverse DNA prompts [15]. Half of the recombinases it generated were active in human cells, at what the authors report as therapeutically relevant levels of CAR insertion in primary human T cells [16]. The percentages arrive without the number of designs in either denominator [22].
Active in human cells means cells in culture. The therapies Basecamp describes would reprogram a patient's cells inside the body, and the company argues that route overcomes the high manufacturing cost and limited complexity of present-day cell therapies [4]. A functional hit rate counts designs that do something in an assay. Where else in the genome an enzyme acts, and for how long the insertion holds, is a different experiment. On the published record EDEN is a sequence-design engine with a measured in-cell success rate, and the in vivo medicine is a plan. Basecamp says it has shown strong preclinical results across multiple modalities and disease areas [8].
Anthropic's money came through the Anthology Fund, a $100 million vehicle launched in 2024 and financed by Menlo Ventures [10]. Companies it backs also get access to Anthropic products and research and $25,000 in credits toward Anthropic's most advanced models [11], about one 5,600th of what Basecamp just raised [19]. Basecamp's own models run inside Anthropic's products: Claude Science, the research workbench launched in June, features its antibiotic design and vaccine prediction models [12]. Both Nvidia and Anthropic were also partners in building the Trillion Gene Atlas, assembled from access and benefit-sharing partnerships in more than 30 countries across all seven continents [18].
The preprint's other demonstration, a generated synthetic microbiome covering 9,067 species at 99 percent biome-specific taxonomic accuracy [17], is a statement about the breadth of that dataset. "We believe the future of medicine lies in reprogramming the body to repair itself. We design the models and the medicines to teach it how," Glen Gowers, Basecamp's co-founder and chief executive, said [6]. Aiming to widen its biopharma partnerships, the company in April appointed Richard Pearce, previously Biogen's head of business development, strategy and portfolio management, as chief business officer [9].
What to watch
- Whether a peer-reviewed version of the January preprint adds denominators, off-target integration data, and any animal in vivo results.
- Whether Basecamp names a biopharma partner under Richard Pearce, which would put an outside price on the EDEN platform.
- Whether Anthropic and Nvidia keep co-authoring the science of a company they fund, or step back to investor-only roles.